ABSTRACT The detection and tracking of individuals in fire environments remain challenging because smoke, flames and time‐varying clutter can mask or distort radar returns. This paper investigates radar‐based recognition of a moving person in a real fire‐test environment using 79 and 300 GHz radar measurements, with emphasis on how frequency, propagation and scene dynamics affect target observability. Range‐Doppler analysis is used to separate moving human returns from stationary fire‐scene reflections, and a region‐based tracking stage is applied to estimate the firefighter trajectory under severe clutter. Experimental results show that 79 GHz radar provides more reliable target visibility and track continuity in the tested scenarios, whereas 300 GHz radar is more limited by lower transmit power and reduced practical detection robustness in the same environment. The study supports the development of radar sensing strategies for firefighter safety and rescue operations in visually degraded environments.

With the rise of commercial constellation implementation in low earth orbit (LEO), the near-Earth space environment is becoming increasingly challenging to monitor and protect. As well as carefully considered policy frameworks, new observational techniques and instrumentation are needed to ensure that safe operations can be maintained by all space users. The Pervasive Sensing group at the University of Birmingham is exploring in-orbit conditional monitoring of satellites using inverse synthetic aperture radar (ISAR) as a technique for dedicated observation of high-value space-based assets. Our previous concept and design results for fixed-beam dual freqency ISAR observations in circular orbits have been extended to a variety of scenarios. I will discuss some of our recent results from both experiments and simulation.

Reverse forward scatter radar is a new method of signal processing that allows for the measurement of forward scattered signals from a monostatic radar that illuminates the target against some background surface reflector. Theoretical analysis shows that measuring the forward scattered instead of back scattered signals significantly improves the signal-to-noise ratio when detecting a stealth target.

The degradation of radar performance due to interference depends on both the interference power and the relationship between the victim and interference waveform parameters. This paper introduces a novel, generalized approach for analyzing the performance of FMCW radar in the presence of interference using a universal graphical tool - the Heatmap. This tool enables: (i) a comprehensive assessment of interference effects on radar performance based on waveform parameters, helping to identify critical cases; (ii) an evaluation of mitigation algorithm effectiveness across a wide range of interference scenarios; and (iii) a straightforward method for estimating signal-to-interference ratio in diverse scenarios. These applications are explored in detail with examples. Furthermore, the proposed calculation method of signal-to-interference ratio is tested through simulations and real experimental data.

The Pervasive Sensing group at the University of Birmingham is exploring in-orbit conditional monitoring of satellites using inverse synthetic aperture radar (ISAR) as a technique for dedicated observation of high-value space-based assets. In this work, the feasibility of geostationary orbit (GEO) observation by optimising monitoring satellite orbital parameters for sub-THz ISAR data acquisition has been assessed. A proprietary propagation simulator, Gofod, has been used to devise the scenarios for which launch conditions, stability, periodicity and time of dwell on the target will deliver the best observation of key observed satellite features. Simulation results have been validated with commercial software.

The fusion of data from different sensing nodes, operating within a distributed sensing suite, is essential for reliable 360◦ imagery around a platform to provide enhanced situational awareness. This work presents a methodology for synchronising data between different sensors and generating multi-modal imagery from a multi-perspective sensing suite, comprising both proprioceptive and exteroceptive sensors.

This paper presents experimental results of the application of Doppler Beam Sharpening (DBS) to enhance the resolution and detectability of maritime targets, with a particular focus on marine infrastructure. Two approaches are investigated: (i) 77 GHz multi-modal sensing based on combined Multiple-Input Multiple-Output (MIMO) and DBS processing, and (ii) 150 GHz real-aperture radar with DBS beamforming. The performance of these beamformers is evaluated and compared with a LiDAR point cloud to highlight the advantages of higher frequencies for next-generation radar sensors.

This paper presents a comprehensive review of advancements in road surface classification technology utilising automotive microwave sensors, covering both active radar and passive radiometry, along with data analysis techniques. Accurate knowledge of road surface type and condition is crucial for improving driving safety, especially in the pursuit of fully autonomous driving. The paper begins with a comparative analysis of different sensing technologies, including microwave, optical, LIDAR and sonar sensors. It subsequently highlights the distinct advantages of microwave sensors, particularly in scenarios with low visibility, where other sensing methods are not sufficiently effective. The analysis of road surface classification methods using radar or radiometer data includes both technical aspects (signal parameters, sensor type, position and number of antennas, signal polarisation, etc.) and classification algorithms. These include analysing backscattered or emitted signal parameters based on specific criteria and making decisions based on this analysis or using statistical classification methods (e.g., k-nearest neighbours, support vector machines, neural networks). The paper also discusses the current state of the field and explores future directions and potential advancements in surface classification technology.

This paper investigates the use of sonar and radar-based sensors for real-time road surface recognition in automotive systems, with the goal of improving road safety. The study demonstrates that statistical analysis of backscattered signals from actual road surfaces, when enhanced by machine learning, can significantly improve the accuracy of surface type classification. The research also shows that fusing data from multiple sensors and increasing radar signal frequency further improve classification accuracy, with an 8% to 12% increase in recognition observed in our experimental data. These findings lay a strong foundation for the development of advanced automotive systems capable of accurately recognizing road surfaces in real time during vehicle motion.

Advancement toward fully autonomous systems requires enhanced sensing and perception, particularly a 360 degrees vision for safe maneuvering. One approach to achieving this is through a distributed network of radar sensors, operating in homogeneous or heterogeneous configurations, strategically positioned to provide increased coverage and visibility in otherwise blind regions. Such a multiperspective sensing network, complemented with multimodal signal processing, can significantly improve the angular resolution of the radar, delivering high-fidelity scene imagery essential for region classification and path planning. This study presents a methodology for multimodal and multiperspective sensing using heterogeneous radar sensors, utilizing Doppler beam sharpening (DBS) within multiple-input-multiple-output (MIMO) radars to enhance the resolution and coverage. Traditional frequency-modulated continuous wave (FMCW)-MIMO radars, currently the most widely used configuration, are prone to Doppler aliasing, limiting the field of view (FoV) in DBS and MIMO-DBS processing. To address this limitation, the effective FoV in multiperspective image is extended to that provided by the radar's physical aperture. The proposed framework is validated using 77-GHz radar chipsets in both automotive and maritime conditions, with sensors mounted in front-looking, corner-looking, and side-looking orientations.

Sub-THz (200 GHz - 700 GHz) Radar has been shown to provide superior imaging capabilities using Inverse Synthetic Aperture Radar (ISAR) for target recognition, identifying the state of an object in the space domain from a space-based sensor. Higher frequencies provide inherently larger bandwidths which achieves a much finer range resolution, higher sensitivity to surface textures which allows fine details of the object to be resolved, as well as shorter integration times for refined cross-range resolutions. Man-made Resident Space Objects (RSO) have a number of deployables (e.g. antennas, robotic arms, probes), which can aid identification of the mission and intent of the RSO, yet they might be very difficult to image due to their low reflectivity towards the observer. Instead, if the shadow cast by the deployable onto the body of the object can be resolved within the image it can serve as a descriptor of such a deployable and therefore inform the observer about the mission and/or state of the object. The capability of resolving shadows in ISAR processing at 300 GHz is presented.

Reverse forward scatter radar is a new approach to monostatic radar where a target is measured against some background reflecting surface and forward scattered signals are measured from the target. A theoretical analysis of the reverse forward scatter radar signal-to-clutter ratio and its relation to the signal-to-noise ratio is made for potential ground-based and airborne monostatic radars.

This paper outlines a method for segmentation and classification of ISAR images generated at Sub-THz frequencies for the purposes of space domain awareness. Image segmentation is achieved using statistical region merging. Simulated ISAR imagery is segmented into simple regions, which are used to train a machine learning model to predict the classes within a series of test images. The results indicate that the use of support vector machines for statistical inference has great potential as part of a broader classification process, able to use multiple predictors to draw distinctions between a number of classes.

This paper describes a methodology of detecting and tracking targets in maritime conditions from a moving radar platform using a combination of MIMO beamformer, CFAR detection, clustering, and a multi-target tracker based on an extended Kalman filter. For validation, experiments have been conducted using compact 77-GHz automotive MIMO radar on a lake with an approaching paddler as the target of interest. The results show a reliable performance with a mean error of 0.32 m in the range and 1.39 degrees the angle, estimated within a 20 s observation interval.

Reverse forward scatter radar (RFSR) is a novel sensor which features a monostatic radar (MR) that illuminates the target against a background reflector. The crucial distinction from any other MR is that the forward scattering (FS) effect is exploited to measure the target using its forward scatter cross section, rather than the monostatic radar cross section (RCS). The first RFSR experimental measurements are shown with verification that the measured amplitude matches the expected values set by previous work on the theoretical power budget. Finally, the application of RFSR to stealth target detection is validated by showing the RFSR signature of a low RCS foam target is the same as that of a relatively higher RCS aluminum (Al) foil target of the same shape and size.

High resolution radar sensing is essential to provide situational awareness to small and medium sized marine platforms. However, detecting small targets on the sea surface is a challenging task for the marine surveillance radars because of the weak echoes and relatively low velocity. While there is a similarity and significant body of research on high resolution radar sensing in automotive environment, the direct translation of such techniques to marine sensing is difficult due to fundamentally dynamic underlaying sea surface. This paper addresses the need of developing novel radar sensing capabilities to image and, potentially, classify small marine targets, such as paddlers, buoys, flotsam and jetsam, or the incoming large waves. Our proposed approach combines Multiple Input, Multiple Output (MIMO) and Doppler Beam Sharpening (DBS) beamforming techniques with the Ordered Statistics – Cell Averaging Constant False Alarm Rate (OSCA-CFAR) for robust target detection, Density Based Spatial Clustering of Applications with Noise (DBSCAN) for clustering, and an adaptive focusing technique. With the developed methodology, multiple small ‘dynamic’ targets within the marine scene have been imaged and detected against substantially suppressed sea background.

Reverse forward scatter radar (RFSR) is a monostatic radar that uses a background reflector to measure the forward scattered signals from a target. A spatially distributed reflector is used in the background for the first time to record the RFSR signature of a target. The results experimentally verify previous theoretical work by matching a measured and simulated signature. The ability of RFSR to measure stealth targets against a spatially distributed background reflector is demonstrated by showing that the RFSR signatures of two different targets are identical. The targets have the same shape and size but significantly differ in the backscatter radar cross section (RCS).

AbstractAn important aspect of Space Situational Awareness is to estimate the intent of objects in space. This paper discusses how discriminating features can be obtained from Inverse Synthetic Aperture Radar images of such objects and how these discriminators can be used to recognise the objects or to estimate their intent. If the object is, for example, a satellite of a known type, the scheme proposed is able to recognise it. The ability of the scheme to detect damage to the object is also discussed. The focus is on imagery obtained in the sub‐terahertz band (typically 300 GHz) because of the greater imaging capability given by the diffuse scattering which is observed at these frequencies. The paper also discusses the importance of being able to use images obtained by electromagnetic simulation to be able to train the subsystem which recognises features of the objects and describes a practical scheme for creating these simulations for large objects at these very short wavelengths.

Mutual interference between radars is one of the major obstacles currently faced by automotive industry towards the attainment of full vehicle autonomy. This paper presents an analysis and mitigation of interference in the spatial domain and evaluates its efficiency compared to the traditional mitigation techniques for various road scenarios and radar configurations. The performance metrics, including peak-to-highest side lobe levels and signal to interference plus noise ratio, have been assessed based on the angular separations between the target and interferers to highlight the cases where spatial domain mitigation might be suited.

In this paper we demonstrate the implementation and validation of a metaheuristic ISAR simulator, capable of generating sub-THz ISAR imagery of space objects for automatic image classification and target recognition. The simulator utilises graphical modelling software Blender and the inverse ray-tracing engine Cycles, using geometric optics approximations to increase computational efficiency. Comparison with results of measurement and simulation in electromagnetic full-wave software have shown that the simulator is able to produce physically meaningful image results with remarkable accuracy, while dramatically reducing computation time.